NOAA Storm Database. Some Explorations and Analysis
Beatriz Valdez
Sunday, May 17, 2015
Synopsis
The NOAA Storm Database has been downloaded from the Reproducible Reasearch Course web site, and uploaded in R using the ‘read.csv’ function. This dataset has been explored to find possible inconsistencies and to select those variables needed to answer the questions that were to guide the analyis. The dataset has been tidied following some of the principles of tidy data: (1) each variable forms a column, (2) each observation forms a row, (3) obvious mistakes in the data are removed, (5) variable values are internally consistent. This process was followed by taking in consideration the events which seemed to be more harmful with respect to population health and to have the greatest economic consequences.
Tornadoes are the most harmful events with respect to the population health, followed by floods, while hurricane/typhoons have had the greatest economic consequences, followed by tornadoes. While many of the weather events in this dataset seems to be associated to some seasons (winter storm, ice storm, wildfire etc.) , tornadoes occur more frequently spanning different seasons during the year.
Data Processing
Loading Data
To load the data, I used the option to read bz connections available in the function read.csv() and stored it in an object: ‘storm’
# loading the data
storm<-read.csv(bzfile(description="~/storm/repdata-data-StormData.csv.bz2",
open="r"), stringsAsFactors=FALSE)I check on the variable names to know which ones I need for my analysis
names(storm)## [1] "STATE__" "BGN_DATE" "BGN_TIME" "TIME_ZONE" "COUNTY"
## [6] "COUNTYNAME" "STATE" "EVTYPE" "BGN_RANGE" "BGN_AZI"
## [11] "BGN_LOCATI" "END_DATE" "END_TIME" "COUNTY_END" "COUNTYENDN"
## [16] "END_RANGE" "END_AZI" "END_LOCATI" "LENGTH" "WIDTH"
## [21] "F" "MAG" "FATALITIES" "INJURIES" "PROPDMG"
## [26] "PROPDMGEXP" "CROPDMG" "CROPDMGEXP" "WFO" "STATEOFFIC"
## [31] "ZONENAMES" "LATITUDE" "LONGITUDE" "LATITUDE_E" "LONGITUDE_"
## [36] "REMARKS" "REFNUM"Selecting Variables
I then select the variables I need. I am going to use packages dplyr, tidyr and lubridates.
# loading packages
require(tidyr)
require(dplyr)
require(lubridate)
# creating a subset from object 'storm'
stormSubset <- storm %>%
select(date = BGN_DATE,
evtype = EVTYPE,
fatalities = FATALITIES,
injuries = INJURIES,
property = PROPDMG,
crop = CROPDMG,
exp1 = PROPDMGEXP,
exp2 = CROPDMGEXP) Basic Explorations
I run some explorations on my subset
# inspecting variable classes
stormSubset %>%
sapply(., function(.) class(.))## date evtype fatalities injuries property crop
## "character" "character" "integer" "integer" "numeric" "numeric"
## exp1 exp2
## "character" "character"# exploring fatalit
stormSubset %>% count(fatalities) %>%
arrange(desc(n)) %>%
top_n(n = 20)## Selecting by n## Source: local data frame [21 x 2]
##
## fatalities n
## 1 0 895323
## 2 1 5010
## 3 2 996
## 4 3 314
## 5 4 166
## 6 5 114
## 7 6 71
## 8 7 53
## 9 8 33
## 10 9 30
## .. ... ...# exploring injuries
stormSubset %>% count(injuries) %>%
arrange(desc(n)) %>%
top_n(n = 20)## Selecting by n## Source: local data frame [20 x 2]
##
## injuries n
## 1 0 884693
## 2 1 7756
## 3 2 3134
## 4 3 1552
## 5 4 931
## 6 5 709
## 7 6 529
## 8 7 280
## 9 10 271
## 10 8 255
## 11 9 186
## 12 12 181
## 13 15 138
## 14 20 130
## 15 11 109
## 16 13 84
## 17 14 84
## 18 30 67
## 19 25 65
## 20 50 58# exploring property damages
stormSubset%>% count(property) %>%
arrange(desc(n)) %>%
top_n(n = 20)## Selecting by n## Source: local data frame [20 x 2]
##
## property n
## 1 0.0 663123
## 2 5.0 32655
## 3 10.0 22018
## 4 1.0 19069
## 5 2.0 17872
## 6 25.0 17696
## 7 50.0 13793
## 8 3.0 10745
## 9 20.0 9307
## 10 15.0 8735
## 11 250.0 8476
## 12 2.5 8220
## 13 0.5 6790
## 14 100.0 6269
## 15 30.0 4443
## 16 500.0 3929
## 17 4.0 3420
## 18 8.0 2967
## 19 75.0 2442
## 20 40.0 2412# exploring exponents related to property damages
stormSubset %>% count(exp1) %>%
arrange(desc(n)) %>%
top_n(n = 20)## Selecting by n## Source: local data frame [19 x 2]
##
## exp1 n
## 1 465934
## 2 K 424665
## 3 M 11330
## 4 0 216
## 5 B 40
## 6 5 28
## 7 1 25
## 8 2 13
## 9 ? 8
## 10 m 7
## 11 H 6
## 12 + 5
## 13 7 5
## 14 3 4
## 15 4 4
## 16 6 4
## 17 - 1
## 18 8 1
## 19 h 1# exploring crop damages
stormSubset %>% count(crop) %>%
arrange(desc(n)) %>%
top_n(n = 20)## Selecting by n## Source: local data frame [20 x 2]
##
## crop n
## 1 0.0 880198
## 2 5.0 4276
## 3 10.0 2381
## 4 50.0 2011
## 5 1.0 1404
## 6 100.0 1237
## 7 2.0 1160
## 8 25.0 846
## 9 20.0 834
## 10 500.0 720
## 11 3.0 655
## 12 15.0 630
## 13 0.5 575
## 14 250.0 515
## 15 200.0 476
## 16 30.0 331
## 17 4.0 301
## 18 75.0 290
## 19 150.0 270
## 20 300.0 249# exploring exponents related to crop damages
stormSubset %>% count(exp2) %>%
arrange(desc(n)) %>%
top_n(n = 20)## Selecting by n## Source: local data frame [9 x 2]
##
## exp2 n
## 1 618413
## 2 K 281832
## 3 M 1994
## 4 k 21
## 5 0 19
## 6 B 9
## 7 ? 7
## 8 2 1
## 9 m 1# exploring even types
stormSubset %>% count(evtype) %>%
arrange(desc(n))%>%
top_n(n = 20)## Selecting by n## Source: local data frame [20 x 2]
##
## evtype n
## 1 HAIL 288661
## 2 TSTM WIND 219940
## 3 THUNDERSTORM WIND 82563
## 4 TORNADO 60652
## 5 FLASH FLOOD 54277
## 6 FLOOD 25326
## 7 THUNDERSTORM WINDS 20843
## 8 HIGH WIND 20212
## 9 LIGHTNING 15754
## 10 HEAVY SNOW 15708
## 11 HEAVY RAIN 11723
## 12 WINTER STORM 11433
## 13 WINTER WEATHER 7026
## 14 FUNNEL CLOUD 6839
## 15 MARINE TSTM WIND 6175
## 16 MARINE THUNDERSTORM WIND 5812
## 17 WATERSPOUT 3796
## 18 STRONG WIND 3566
## 19 URBAN/SML STREAM FLD 3392
## 20 WILDFIRE 2761# exploring date
stormSubset %>% count(date) %>%
arrange(desc(n))%>%
top_n(n = 20)## Selecting by n## Source: local data frame [20 x 2]
##
## date n
## 1 5/25/2011 0:00:00 1202
## 2 4/27/2011 0:00:00 1193
## 3 6/9/2011 0:00:00 1030
## 4 5/30/2004 0:00:00 1016
## 5 4/4/2011 0:00:00 1009
## 6 4/2/2006 0:00:00 981
## 7 4/7/2006 0:00:00 973
## 8 4/19/2011 0:00:00 970
## 9 5/31/1998 0:00:00 933
## 10 5/22/2011 0:00:00 919
## 11 6/4/2008 0:00:00 900
## 12 5/21/2004 0:00:00 899
## 13 5/26/2011 0:00:00 885
## 14 3/12/2006 0:00:00 883
## 15 6/21/2011 0:00:00 876
## 16 5/6/2003 0:00:00 830
## 17 4/3/2007 0:00:00 801
## 18 5/24/2004 0:00:00 798
## 19 6/10/2008 0:00:00 782
## 20 5/25/2008 0:00:00 779Basic Data Munging
I need to perform some basic data munging
Each variable forms a column
Categories ‘exp1’ and ‘exp2’ have to be gather under one variable and their values have to be stored in another one. First, I need to normalize them. I also require package ‘magrittr’
# loading magrittr
require(magrittr)
# gathering 'exp1' and 'exp2'
stormSubset %<>%
mutate(exp1 = chartr('m', 'M', exp1),
exp2 = chartr('k', 'K', exp2),
exp2 = chartr('m', 'M', exp2)) %>%
gather(exp, amount, exp1:exp2)
# inspecting
glimpse(stormSubset)## Observations: 1804594
## Variables:
## $ date (chr) "4/18/1950 0:00:00", "4/18/1950 0:00:00", "2/20/195...
## $ evtype (chr) "TORNADO", "TORNADO", "TORNADO", "TORNADO", "TORNAD...
## $ fatalities (int) 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 4, 0, ...
## $ injuries (int) 15, 0, 2, 2, 2, 6, 1, 0, 14, 0, 3, 3, 26, 12, 6, 50...
## $ property (dbl) 25.0, 2.5, 25.0, 2.5, 2.5, 2.5, 2.5, 2.5, 25.0, 25....
## $ crop (dbl) 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...
## $ exp (fctr) exp1, exp1, exp1, exp1, exp1, exp1, exp1, exp1, ex...
## $ amount (chr) "K", "K", "K", "K", "K", "K", "K", "K", "K", "K", "...stormSubset %>% count(amount) %>%
arrange(desc(n))%>%
top_n(n = 20)## Source: local data frame [18 x 2]
##
## amount n
## 1 1084347
## 2 K 706518
## 3 M 13332
## 4 0 235
## 5 B 49
## 6 5 28
## 7 1 25
## 8 ? 15
## 9 2 14
## 10 H 6
## 11 + 5
## 12 7 5
## 13 3 4
## 14 4 4
## 15 6 4
## 16 - 1
## 17 8 1
## 18 h 1# filtering out values different to 'k', 'm' and 'b'
stormSubset %<>%
filter(amount == 'K' | amount == 'B' | amount == 'M')
#inspecting
stormSubset %>% count(amount) %>%
arrange(desc(n))%>%
top_n(n = 20)## Source: local data frame [3 x 2]
##
## amount n
## 1 K 706518
## 2 M 13332
## 3 B 49# inspecting
glimpse(stormSubset)## Observations: 719899
## Variables:
## $ date (chr) "4/18/1950 0:00:00", "4/18/1950 0:00:00", "2/20/195...
## $ evtype (chr) "TORNADO", "TORNADO", "TORNADO", "TORNADO", "TORNAD...
## $ fatalities (int) 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 4, 0, ...
## $ injuries (int) 15, 0, 2, 2, 2, 6, 1, 0, 14, 0, 3, 3, 26, 12, 6, 50...
## $ property (dbl) 25.0, 2.5, 25.0, 2.5, 2.5, 2.5, 2.5, 2.5, 25.0, 25....
## $ crop (dbl) 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, ...
## $ exp (fctr) exp1, exp1, exp1, exp1, exp1, exp1, exp1, exp1, ex...
## $ amount (chr) "K", "K", "K", "K", "K", "K", "K", "K", "K", "K", "...Fatalities and injuries are two categories of one single variable. Let’s gather them together and filter out zero damage cots. I also need to translate amount characters into their numeric counterpart
stormSubset %<>%
gather(damage, cost, property:crop) %>%
filter(cost > 0)
stormSubset %<>%
mutate (amount = gsub('K',
'1000', amount),
amount = gsub('M',
'1000000', amount),
amount = gsub('B',
'1000000000', amount))
# inspecting
glimpse(stormSubset)## Observations: 372852
## Variables:
## $ date (chr) "4/18/1950 0:00:00", "4/18/1950 0:00:00", "2/20/195...
## $ evtype (chr) "TORNADO", "TORNADO", "TORNADO", "TORNADO", "TORNAD...
## $ fatalities (int) 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 4, 0, 0, ...
## $ injuries (int) 15, 0, 2, 2, 2, 6, 1, 0, 14, 0, 3, 3, 26, 6, 50, 2,...
## $ exp (fctr) exp1, exp1, exp1, exp1, exp1, exp1, exp1, exp1, ex...
## $ amount (chr) "1000", "1000", "1000", "1000", "1000", "1000", "10...
## $ damage (fctr) property, property, property, property, property, ...
## $ cost (dbl) 25.0, 2.5, 25.0, 2.5, 2.5, 2.5, 2.5, 2.5, 25.0, 25....# transforming variable 'amount' into a numeric one
stormSubset$amount <-as.numeric(stormSubset$amount)
class(stormSubset$amount)## [1] "numeric"# multiplying 'cost' * 'amount' and tranforming the cost into billions
# eliminating then variable 'amount' from my dataset
stormSubset %<>%
mutate(cost = amount * cost,
cost=cost/10^9 ) %>%
select(-amount)Gathering no ‘fatalities’, and ‘injuries’ under a common variable, and creating a new one to store their values
# gathering and filtering out harm
stormSubset %<>%
gather(harm, harmImpact,
fatalities:injuries) %>%
filter(harmImpact > 0)
# inspecting
glimpse(stormSubset)## Observations: 18290
## Variables:
## $ date (chr) "2/13/1952 0:00:00", "2/13/1952 0:00:00", "3/22/195...
## $ evtype (chr) "TORNADO", "TORNADO", "TORNADO", "TORNADO", "TORNAD...
## $ exp (fctr) exp1, exp1, exp1, exp1, exp1, exp1, exp1, exp1, ex...
## $ damage (fctr) property, property, property, property, property, ...
## $ cost (dbl) 2.5e-05, 2.5e-04, 2.5e-05, 2.5e-05, 2.5e-03, 2.5e-0...
## $ harm (fctr) fatalities, fatalities, fatalities, fatalities, fa...
## $ harmImpact (int) 1, 1, 4, 1, 6, 7, 2, 5, 25, 2, 4, 1, 3, 1, 2, 1, 10...Fixing Variable ‘date’
I now fix the varible date
# getting rid of hms:
dtes <-strsplit(as.character(stormSubset$date), " ")
#mutating variable date and subsetting to get data from 1990 up
# and changing cases within evtype categories
stormSubset %<>%
mutate(date=sapply(dtes, '[', 1))
stormSubset %<>% filter(mdy(date) > mdy('1/1/1990')) %>%
mutate(date = mdy(date), evtype = tolower(evtype))
glimpse(stormSubset)## Observations: 12483
## Variables:
## $ date (time) 1991-03-29, 1992-03-10, 1990-03-14, 1992-10-03, 19...
## $ evtype (chr) "tornado", "tornado", "tornado", "tornado", "tornad...
## $ exp (fctr) exp1, exp1, exp1, exp1, exp1, exp1, exp1, exp1, ex...
## $ damage (fctr) property, property, property, property, property, ...
## $ cost (dbl) 0.00025, 0.00025, 0.00025, 0.00250, 0.02500, 0.0025...
## $ harm (fctr) fatalities, fatalities, fatalities, fatalities, fa...
## $ harmImpact (int) 5, 2, 1, 1, 3, 1, 4, 1, 1, 29, 1, 1, 6, 1, 1, 1, 4,...Fixing Category Names
Each variable form a column and each observation form a row. Now we need to fix some categories within “evtype” variable. Let’s to have a visual approach of these needed clean using explortory graphs
library(ggplot2)
library(gridExtra)## Loading required package: gridg1<-qplot(yday(date), evtype,
data=subset(stormSubset, harmImpact > 50), size=harmImpact) +
facet_wrap(~harm)
g2<-qplot(yday(date), evtype,
data=subset(stormSubset, cost > 5.0e-4), size=cost) +
facet_wrap(~damage)
grid.arrange(g1, g2, ncol=2, main="Searching malformed names")
# inspecting evtype categories
stormSubset %>% count(evtype) %>%
arrange(desc(n))%>%
top_n(n = 20)## Selecting by n## Source: local data frame [20 x 2]
##
## evtype n
## 1 tornado 4242
## 2 tstm wind 1468
## 3 flash flood 1128
## 4 thunderstorm wind 1047
## 5 flood 697
## 6 high wind 575
## 7 lightning 469
## 8 thunderstorm winds 369
## 9 wildfire 248
## 10 hail 237
## 11 strong wind 232
## 12 winter storm 177
## 13 heavy snow 139
## 14 ice storm 101
## 15 blizzard 87
## 16 wild/forest fire 87
## 17 tropical storm 86
## 18 heavy rain 84
## 19 hurricane 81
## 20 high winds 73# checking for extra spaces
grep('[ \t]{2,}',perl=TRUE, value=TRUE,stormSubset$evtype)## character(0)grep('\\t+\\s+$',perl=TRUE, value=TRUE, stormSubset$evtype)## character(0)# tstm wind into thunderstorm wind
stormSubset$evtype <-gsub('tstm wind', 'thunderstorm wind',
stormSubset$evtype)
# into 'thunderstorm wind'
stormSubset$evtype <-gsub(' thunderstorm wind)', 'thunderstorm wind',
stormSubset$evtype)
# into 'thunderstorm wind'
stormSubset$evtype <-gsub('^thunderstorm wind/\\w+)', 'thunderstorm wind',
stormSubset$evtype)
# thunderstorm winds into thunderstorm wind
stormSubset$evtype <-gsub('thunderstorm winds', 'thunderstorm wind',
stormSubset$evtype)
# any '^thunderstorm wind\\s+' into 'thunderstorm wind'
stormSubset$evtype <-gsub('^thunderstorm wind\\s+', 'thunderstorm wind',
perl=TRUE,
stormSubset$evtype)
# '^thunderstorm\\s+wind\\(\\w+' into 'thunderstorm wind'
stormSubset$evtype <-gsub('^thunderstorm\\s+wind\\(\\w+',
'thunderstorm wind',
perl=TRUE,
stormSubset$evtype)
stormSubset$evtype <-gsub('^thunderstorm\\s+\\w+/',
'thunderstorm wind',
perl=TRUE,
stormSubset$evtype)
# winter weather/mix into winter weather
stormSubset$evtype <-gsub('winter storms', ' winter storm',
stormSubset$evtype)
# winter storms into winter storm
stormSubset$evtype <-gsub('winter storms', ' winter storm',
stormSubset$evtype)
# strong winds into strong wind
stormSubset$evtype <-gsub('strong wind', 'strong winds',
stormSubset$evtype)
# 'high winds' into 'high wind'
stormSubset$evtype <-gsub('high winds', 'high wind',
stormSubset$evtype)
# any wild\\s+fire\\w+' into 'wildfire'
stormSubset$evtype <-gsub('wild\\s+fire\\w+', 'wildfire',
perl=TRUE,
stormSubset$evtype)
#
stormSubset$evtype <-gsub('wild/\\w+', 'wildfire',
perl=TRUE,
stormSubset$evtype)
#
stormSubset$evtype <-gsub('wildfire\\s+\\w+', 'wildfire',
perl=TRUE,
stormSubset$evtype)
# any ^coastal\\s+\\w+' into 'coastal flood'
stormSubset$evtype <-gsub('^coastal\\s+\\w+',
'coastal flood',
perl=TRUE,
stormSubset$evtype)
# any ^coastal\\s+\\w+'\\w+ into 'coastal flood'
stormSubset$evtype <-gsub('^coastal\\s+\\w+/\\w+',
'coastal flood',
perl=TRUE,
stormSubset$evtype)
#any ^tornadoes\\s+\\w+' into 'tornado'
stormSubset$evtype <-gsub('^tornadoes,\\s+\\w+',
'tornado',
perl=TRUE,
stormSubset$evtype)
#any ^tornado\\s+\\w+' into 'tornado'
stormSubset$evtype <-gsub('^tornado\\s+\\w+',
'tornado',
perl=TRUE,
stormSubset$evtype)
#any '^tornado\\s+\\w+' into 'tornado'
stormSubset$evtype <-gsub('^tornado\\s+\\w+',
'tornado',
perl=TRUE,
stormSubset$evtype)
#any '^tornado,\\s+\\w+' into 'tornado'
stormSubset$evtype <-gsub('^tornado,\\s+\\w+',
'tornado',
perl=TRUE,
stormSubset$evtype)
#any '^storm\\s+\\w+' into 'storm surge'
stormSubset$evtype <-gsub('^storm\\s+\\w+/',
'storm surge',
perl=TRUE,
stormSubset$evtype)
#any '^hurricane$' into 'hurricane/typhoon'
stormSubset$evtype <-gsub('^hurricane$',
'hurricane/typhoon',
perl=TRUE,
stormSubset$evtype)
#any '^heat\\s+\\w+' into 'heat'
stormSubset$evtype <-gsub('^heat\\s+\\w+',
'heat',
perl=TRUE,
stormSubset$evtype)
#
stormSubset$evtype <-gsub('^heat\\s+\\w+',
'heat',
perl=TRUE,
stormSubset$evtype)
# transforming flash flooding, flash flash flood etc.
stormSubset$evtype <-gsub('flash flooding',
'flash flood',
perl=TRUE,
stormSubset$evtype)
#
stormSubset$evtype <-gsub('(flash flood/|flood/flash)',
'flash flood',
perl=TRUE,
stormSubset$evtype)
#
stormSubset$evtype <-gsub('\\w+flood\\.?',
'flash flood',
perl=TRUE,
stormSubset$evtype)
#
stormSubset$evtype <-gsub('flash flash flood',
'flash flood',
perl=TRUE,
stormSubset$evtype)
#
stormSubset$evtype <-gsub('flash flood flood',
'flash flood',
perl=TRUE,
stormSubset$evtype)
#
stormSubset$evtype <-gsub('flooding',
'flood',
perl=TRUE,
stormSubset$evtype)
# (storm surgetide|storm surge) into 'storm surge/tide'
stormSubset$evtype <-gsub('(storm surgetide|storm surge)',
'storm surge/tide',
perl=TRUE,
stormSubset$evtype)
# winter weather/' into 'winter weather'
stormSubset$evtype <-gsub('winter weather/',
'winter weather',
perl=TRUE,
stormSubset$evtype)
# 'winter weathermix' 'winter weather'
stormSubset$evtype <-gsub('winter weathermix',
'winter weather',
perl=TRUE,
stormSubset$evtype)
# 'winter storm\\s+' into 'winter storm'
stormSubset$evtype <-gsub('winter storm\\s+',
'winter storm',
perl=TRUE,
stormSubset$evtype)
#
stormSubset$evtype <-gsub('winter stormhigh',
'winter storm',
perl=TRUE,
stormSubset$evtype)
# 'ice$' into 'ice storm'
stormSubset$evtype <-gsub('ice$',
'ice storm',
perl=TRUE,
stormSubset$evtype)
# 'smal hail' into 'hail'
stormSubset$evtype <-gsub('small hail',
'hail',
perl=TRUE,
stormSubset$evtype)
# hurricane/typhoons
stormSubset$evtype <-gsub('^hurricane\\s+\\w+',
'hurricane/typhoon',
perl=TRUE,
stormSubset$evtype)
# '^fog$' into 'dense fog'
stormSubset$evtype <-gsub('^fog$',
'dense fog',
perl=TRUE,
stormSubset$evtype)·Reshaping dataset
stormDf <- stormSubset %>%
group_by(evtype, harm) %>%
summarize(harmImpact=sum(harmImpact)) %>%
filter(harmImpact > 50) %>%
arrange(desc(harmImpact))
stormDF <- stormSubset %>%
group_by(evtype, damage) %>%
summarize(costs=sum(cost)) %>%
filter(costs > 5.0e-4) %>%
arrange(desc(costs))
dF <- stormSubset %>%
group_by(date,evtype, harm, damage) %>%
summarize(victims = sum(harmImpact), costs=sum(cost)) %>%
filter(victims > 50, costs > 5.0e-4)
dF %>% count(evtype) %>%
arrange(desc(n)) %>%
top_n(n = 20) ## Source: local data frame [20 x 2]
##
## evtype n
## 1 tornado 143
## 2 flood 12
## 3 flash flood 10
## 4 thunderstorm wind 10
## 5 wildfire 6
## 6 hurricane/typhoon 5
## 7 high wind 4
## 8 ice storm 4
## 9 tropical storm 4
## 10 tsunami 4
## 11 blizzard 3
## 12 heat 3
## 13 winter storm 3
## 14 dense fog 2
## 15 excessive heat 2
## 16 hail 2
## 17 heavy snow 2
## 18 tropical storm gordon 2
## 19 extreme cold 1
## 20 winter weather 1Results
In general, accross the United States, tornadoes have been the most harmful weather event with respect to population health. They have caused many much fatalities and injuries than any other event. Tornadoes are also the second type of weather event which have the greatest economic consequence. Hurricane/typhoons are the first one
g1<-ggplot(data=stormDf, aes(harmImpact, reorder(evtype, harmImpact)))+
geom_point() +
facet_wrap(~harm) +
ggtitle('Type of Event by Number of Victims') +
xlab('Number of Victims') + ylab('Type of Event')
g2<-ggplot(data=stormDF, aes(costs, reorder(evtype, costs))) +
geom_point() +
facet_wrap(~damage) +
ggtitle('Type of Event by Damage (billions of dollars)') +
xlab('day of the year') + ylab('type of event') +
labs(size='Costs\n(billions of Dollars)')
grid.arrange(g1, g2, ncol=1, main="America. Weather Events")
Finally, while other weather events seem to be associated with certain seasons of the year, tornadoes occur in almost any of them:
g3<-ggplot(data=dF, aes(yday(date), evtype,size=victims))+
geom_point() +
facet_grid(harm ~ damage, scales='free_y')+
ggtitle('Event type by date and harm impact') +
xlab('day of the year') + ylab('type of event') +
labs(size='Number of\nVictims')
g4<-ggplot(data=dF, aes(yday(date), evtype,size=costs)) +
geom_point() +
facet_grid(harm ~ damage, scales='free_y') +
ggtitle('Event type by date and damage (in billions of dollars)') +
xlab('day of the year') + ylab('type of event') +
labs(size='Costs\n(billions of Dollars)')
grid.arrange(g3, g4, ncol=1, main="Weather Events")